• Journal of Electrical Engineering and Technology
• /
• v.12 no.1
• /
• pp.363-372
• /
• 2017

This paper presents a novel face recognition algorithm based on the deep convolution neural network and key point detection jointed local binary pattern methodology to enhance the accuracy of face recognition. We firstly propose the modified face key feature point location detection method to enhance the traditional localization algorithm to better pre-process the original face images. We put forward the grey information and the color information with combination of a composite model of local information. Then, we optimize the multi-layer network structure deep learning algorithm using the Fisher criterion as reference to adjust the network structure more accurately. Furthermore, we modify the local binary pattern texture description operator and combine it with the neural network to overcome drawbacks that deep neural network could not learn to face image and the local characteristics. Simulation results demonstrate that the proposed algorithm obtains stronger robustness and feasibility compared with the other state-of-the-art algorithms. The proposed algorithm also provides the novel paradigm for the application of deep learning in the field of face recognition which sets the milestone for further research.

• Journal of Biosystems Engineering
• /
• v.23 no.6
• /
• pp.621-628
• /
• 1998

In order to develop an inspection algorithm for an automatic eggshell inspection system, acoustic impulse response with neural network method was studied. An improved error backpropagation algorithm was selected as a loaming rule of neural network, and three layer network was chosen for the neural network architecture. Acoustic signals in time domain and theirs power spectrum were studied as the input to the neural network. The classification feasibility and success rate were investigated in terms of statistical analysis and neural network approach. As a result, the success rate was 95% with the statistical model having five independent variables. Among the neural network models studied, the power spectrum of acoustic signal as the input with 64 input neurons and the two impact data showed the success rate of 95.5% which was slightly higher than of statistical analysis.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.56 no.8
• /
• pp.1351-1359
• /
• 2007

In this paper, we have studied the power quality(PQ) diagnosis system with the two methods for PQ diagnosis. One to Apply a regulation value in compliance with mathematics calculation, and the other Automatic identification using Neural network algorithm. Neural network algorithm is used for an automatic diagnosis of the PQ. The regulation proposed by IEEE 1159 Working group is applied for the precision of the diagnosis. In order to divide accurate segmentation, the algorithm for a computer training used the back propagation out of several neural network algorithms. We have configured the proto-type sample by using Labview and a programmed Neural Networks Algorithm using with C. And arbitrary electric Signal generated by OMICRON Company's CMC 256-6 for an efficiency test.

• Proceedings of the KIEE Conference
• /
• 1999.07c
• /
• pp.1245-1247
• /
• 1999

This paper describes the optimal reconfiguration of distribution network. The optimal routing of distribution network should provide electricity to customers with quality, and this paper shows that optimal routing of distribution network can be obtained by Neural-Tabu algorithm while keeping constraints such as line power capacity, voltage drop and reliability indices. The Neural-Tabu algorithm is a Tabu algorithm combined with Neural network to find neighborhood solutions. This paper shows that not only the loss cost but also the reliability cost should be considered in distribution network reconfiguration to achieve the optimal routing.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.10
• /
• pp.1968-1975
• /
• 2009

The paper proposes a novel control algorithm for tracking maximum power of PV generation system.. The maximum power of PV array is determinated by a insolation and temperature. Prior considered the term in PV generation system is how maximum power point(MPP) is accurately tracked.. The paper proposes a fuzzy neural network(FNN) control algorithm so as to accurately track those maximum power points. The proposed control algorithm comprises the antecedence part of fuzzy rule and clustering method, multi-layer neural network in the consequent part. FNN has the advantages which are depicted both high performance and robustness in fuzzy control and high adaptive control in neural network.. Specially, it can show the outstanding control performance for parameter variations appling to non-linear character of PV array. In this paper, the tracking speed and the accuracy prove the validity through comparing a proposed algorithm with a conventional one.

• Proceedings of the KIEE Conference
• /
• 2004.07d
• /
• pp.2580-2582
• /
• 2004

In this paper, a Neural Network(NN) based approach for classification of two dimensions images. The proposed algorithm is able to apply in the actual industry. The described diagnostic algorithm is presented to defect surface failures on tiles. A way to get data for a digital image process is several kinds of it. The tiles are scanned and the digital images are preprocessed and classified using neural networks. It is important to reduce the amount of input data with problem specific preprocessing. The auto-associative neural network is used for feature generation and selection while the probabilistic neural network is used for classification. The proposed algorithm is evaluated experimentally using one hundred of the real tile images. Sample image data to preprocess have histogram. The histogram is used as input value of probabilistic neural network. Auto-associative neural network compress input data and compressed data is classified using probabilistic neural network. Classified sample images are determined by human state. So it is intervened human subjectivity. But digital image processing and neural network are better than human classification ability. Therefore it is very useful of quality control improvement.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.64 no.10
• /
• pp.1479-1485
• /
• 2015

Robot's technology was very simple and repetitive in the past. Nowadays, robots are required to perform intelligent operation. So, path planning has been studied extensively to create a path from start position to the goal position. In this paper, potential field algorithm was used for path planning in dynamic environments. It is used for a path plan of mobile robot because it is elegant mathematical analysis and simplicity. However, there are some problems. The problems are collision risk, avoidance path, time attrition. In order to resolve path problems, we amalgamated potential field algorithm with the artificial neural network system. The input of the neural network system is set using relative velocity and location between the robot and the obstacle. The output of the neural network system is used for the weighting factor of the repulsive potential function. The potential field algorithm problem of mobile robot's path planning can be improved by using artificial neural network system. The suggested algorithm was verified by simulations in various dynamic environments.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.35C no.5
• /
• pp.59-68
• /
• 1998

This paper proposes a neural network algorithm for a channel assignment in cellular mobile communications. The proposed algorithm is developed base on hopfield neural network in order to minimize the number of channel without a confliction between cells. To compare the performance of the proposed algorithm, we used seven benchmark problems selected from kunz's and funabiki's papers. Experimental results show that the convergence times are reduced form 27% to 66% compared with Kunz's and funabiki's algorithm and vonvergence rates are improved to 100%.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.7
• /
• pp.81-90
• /
• 2018

In this paper, reducing lense Vignetting effect and adaptive learning rate method are proposed to complement Scribner's neural network for nuc algorithm which is the effective algorithm in statistic SBNUC algorithm. Proposed reducing vignetting effect method is updated weight and bias each differently using different cost function. Proposed adaptive learning rate for updating weight and bias is using sobel edge detection method, which has good result for boundary condition of image. The ordinary statistic SBNUC algorithm has problem to compensate lense vignetting effect, because statistic algorithm is updated weight and bias by using gradient descent method, so it should not be effective for global weight problem same like, lense vignetting effect. We employ the proposed methods to Scribner's neural network method(NNM) and Torres's reducing ghosting correction for neural network nuc algorithm(improved NNM), and apply it to real-infrared detector image stream. The result of proposed algorithm shows that it has 10dB higher PSNR and 1.5 times faster convergence speed then the improved NNM Algorithm.

• Journal of Korea Multimedia Society
• /
• v.23 no.6
• /
• pp.756-764
• /
• 2020

A typical algorithm used for image analysis is the Convolutional Neural Network(CNN). R-CNN, Fast R-CNN, Faster R-CNN, etc. have been studied to improve the performance of the CNN, but they essentially require large amounts of data and high algorithmic complexity., making them inappropriate for small and medium-sized services. Therefore, in this paper, the image label prediction algorithm based on CNN with collaborative layer with low complexity, high accuracy, and small amount of data was proposed. The proposed algorithm was designed to replace the part of the neural network that is performed to predict the final label in the existing deep learning algorithm by implementing collaborative filtering as a layer. It is expected that the proposed algorithm can contribute greatly to small and medium-sized content services that is unsuitable to apply the existing deep learning algorithm with high complexity and high server cost.